Archive verified research artifact: summary.json

artifacts/experiments/runs/qwen3-mmqa_ti-s17/summary.json

@@ -0,0 +1,75 @@
+{
+  "decoder_trainable": false,
+  "scores": [
+    {
+      "complete": true,
+      "dataset": "mmqa_ti",
+      "groups": {
+        "modality:image": {
+          "em": 19.745222929936304,
+          "f1": 22.70063694267516,
+          "n": 314
+        },
+        "modality:image+text": {
+          "em": 13.68421052631579,
+          "f1": 17.178947368421053,
+          "n": 95
+        },
+        "modality:text": {
+          "em": 26.829268292682926,
+          "f1": 32.77591463414634,
+          "n": 656
+        },
+        "type:Compare(Compose(TableQ,ImageQ),Compose(TableQ,TextQ))": {
+          "em": 17.647058823529413,
+          "f1": 20.58823529411765,
+          "n": 17
+        },
+        "type:Compose(ImageQ,TextQ)": {
+          "em": 32.0,
+          "f1": 32.0,
+          "n": 25
+        },
+        "type:Compose(TextQ,ImageListQ)": {
+          "em": 4.0,
+          "f1": 8.3,
+          "n": 50
+        },
+        "type:ImageListQ": {
+          "em": 0.78125,
+          "f1": 6.0703125,
+          "n": 128
+        },
+        "type:ImageQ": {
+          "em": 32.795698924731184,
+          "f1": 34.145161290322584,
+          "n": 186
+        },
+        "type:Intersect(ImageListQ,TextQ)": {
+          "em": 0.0,
+          "f1": 22.333333333333332,
+          "n": 3
+        },
+        "type:TextQ": {
+          "em": 26.829268292682926,
+          "f1": 32.77591463414634,
+          "n": 656
+        }
+      },
+      "metric_scale": "0-100",
+      "n_expected": 1065,
+      "n_missing": 0,
+      "n_predictions": 1065,
+      "parse_success_count": 1054,
+      "scores": {
+        "em": 23.568075117370892,
+        "f1": 28.414084507042254
+      },
+      "status_counts": {
+        "ok": 1065
+      }
+    }
+  ],
+  "seed_is_not_an_independent_train": true,
+  "trainable_method": false
+}